context's flow_id is None in Environment execute()

Description

During testing of DaskCloudProviderEnvironment we noticed that prefect.context.get("flow_id") returns None in our execute() method. prefect.context.get("flow_run_id") does correctly return the flow run id and we are able to retrieve the flow_id with essentially client.get_flow_run_info(flow_run_id).flow_id

Expected Behavior

flow_id should be available in context inside an Environment’s execute() method.

Reproduction

def execute(self, storage: "Storage", flow_location: str, **kwargs: Any) -> None:
    flow_id = prefect.context.get("flow_id")
    if not flow_id:  # We've observed cases where flow_id is None
        if not flow_run_info:
            client = Client()
            flow_run_info = client.get_flow_run_info(flow_run_id)
        flow_id = flow_run_info.flow_id

Environment

Docker storage with the Fargate agent. (Agent is running Fargate tasks for the flow runs.)

[joeschmid]

Ended up making a small, separate PR for this: https://github.com/PrefectHQ/prefect/pull/2461 since it could use a bit of explanation…

For the Docker, Fargate, Local, & Nomad agents it was extremely simple to add this. For the Kubernetes agent and FargateTaskEnvironment this wasn’t readily available. We could obviously add a GraphQL query for these if we want to keep the behavior uniform, i.e. for all agents/environments to make flow_id available in context. Thanks to @cicdw that PR now supports all agents.

[joeschmid]

Hi @lauralorenz, that makes total sense. I’ll plan to include it! Given that I’m in there anyway it’ll be easy to test. (And I’ll back out my funky logic to workaround not having flow_id.)